Acoustic Anomaly Diagnosis Using STFT Temporal Spectral Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anomaly diagnosis technologies face challenges in determining the type of anomaly when different anomalies exhibit the same spectrum pattern, and in diagnosing unknown anomalies that differ from typical anomalies sensed by operators.
Innovation Solution
An anomaly diagnosis device that includes a microphone, a signal converter, and a signal processing device. The signal processing device performs short-time fast Fourier transform on waveform data, calculates feature quantities representing the degree of non-uniformity in temporal change of spectral intensity, and compares these features with stored data of known normal waveforms to determine acceptability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If only spectrum pattern is used for anomaly determination, then the diagnosis method is simple, but the ability to determine the type of anomaly is insufficient when different anomalies show the same spectrum pattern
Solution Approach 1:
The patent transitions from one-dimensional spectrum pattern analysis to multi-dimensional analysis by extracting multiple feature quantities including temporal change characteristics, spectral centroid, spectral rolloff, and zero-crossing rate. This dimensional expansion enables differentiation between anomalies with identical spectrum patterns but different temporal behaviors.
Solution Approach 2:
The patent changes the parameters used for anomaly detection from单一的spectrum pattern to multiple feature quantities representing different aspects of the sound signal. By calculating feature quantities such as temporal change degree, spectral centroid, and zero-crossing rate, the system achieves more precise anomaly classification.
2Reliability
If traditional spectrum pattern comparison is used, then known anomalies can be detected, but unknown anomalies different from usual anomalies cannot be determined
Solution Approach 1:
The patent performs preliminary extraction of multiple feature quantities from the sound signal before comparison with reference data. By pre-calculating comprehensive features including temporal characteristics and spectral properties, the system prepares a robust feature vector that can match both known and unknown anomaly patterns.
Solution Approach 2:
The patent creates a universal diagnosis system that handles both known and unknown anomalies through multi-functional feature extraction. The extracted feature quantities serve multiple purposes: identifying known anomaly types while also detecting unknown anomalies by comparing against multiple reference patterns, making the system adaptable to various anomaly scenarios.
3Device complexity
If auditory sense or tactile sense of operator is used for sensory test, then the test can be performed with simple equipment, but the information concerning vibration or operation sound cannot be quantified
Solution Approach 1:
The patent replaces the mechanical/sensory system of human operators with an electronic signal processing system. By using a microphone to convert sound into electrical signals and then processing these signals computationally to extract feature quantities, the system quantifies vibration and operation sound information that would otherwise remain subjective and unmeasurable.
Solution Approach 2:
The patent introduces an intermediary signal processing system between the sound source and the analysis. The microphone and signal processing unit act as intermediaries that convert physical sound waves into quantifiable electrical signals and feature quantities, bridging the gap between physical vibration and measurable data.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The device effectively determines both known and unknown anomalies, improving the accuracy of anomaly diagnosis by analyzing the degree of discrepancy in feature quantities between input signals and known normal waveforms.
Implementation Method 1
a microphone that converts a sound from a determination target into an analog electrical signal
Implementation Method 2
a signal converter that converts the analog electrical signal into a digital signal
Implementation Method 3
The signal processing unit performs short-time fast Fourier transform on waveform data of an input signal, and calculates feature quantities
Data Source
AI summary
An anomaly diagnosis device includes a signal processing device. The signal processing device includes a signal processing unit, a data storage unit, and a determination unit. The signal processing unit performs STFT on waveform data of an input signal, and calculates feature quantities. The data storage unit stores feature quantity data based on data of a known normal waveform. The determination unit compares first feature quantity data consisting of multiple ones of the feature quantities calculated by the signal processing unit with second feature quantity data, which is the feature quantity data stored in the data storage unit, and determines acceptability of the waveform data of the input signal. The feature quantities are each a quantity representing the degree of non-uniformity in temporal change of spectral intensity of a specific frequency band included in waveform data.